FAD: Foundation-Model Aided Deep Learning for Marine Seismic Interference Attenuation
刘马
GeoBrain · PKU & HIT
Making geophysicsmore intelligent — and trustworthy.
We connect artificial intelligence, applied mathematics and exploration geophysics to build reproducible, interpretable and transferable methods for subsurface imaging and seismic processing.
Open benchmarks and reproducible research
Start here · 新成员必读
Geophysics & AI Starter Handbook
A 78-page onboarding path for new graduate and undergraduate members, covering geophysical tasks, AI methods, research tools and learning resources. This site always points to the latest uploaded edition.
88Methods
44Benchmarks
368Papers
494Results
Research map
Five connected research paths
From raw seismic records to subsurface structures, foundation models and executable geophysical agents.
01
Seismic Data Processing
We study the full chain from missing-trace reconstruction to first-arrival picking under irregular acquisition, random noise and coherent interference. Generative models, self-supervised learning, implicit representations and geometry-aware networks are combined with seismic structure, while SeismicBench records datasets, metrics and reproducible experiments.
02Geophysical Inversion
For velocity-model building, full-waveform inversion and multiphysics problems, wave equations, sensing priors and deep networks are placed in a common optimization framework. Work spans end-to-end inversion, physics-informed generative learning, neural implicit representations and parameter-efficient adaptation, with attention to missing low frequencies, initial-model dependence, 3D cost and uncertainty.
03Artificial Intelligence Algorithms
We derive broadly useful AI methods from geophysical problems, including domain adaptation, noise-robust losses, personalized federated learning, generative modeling and efficient 3D vision. The goal is to explain why models transfer, when they fail and how they can be deployed reliably under limited data and compute.
04Foundation Models for Geophysics
We study scalable pretraining and cross-modal alignment across seismic, gravity, magnetic, electromagnetic and textual knowledge, building foundation models that adapt through prompts, lightweight decoders or low-rank updates. The program also covers data governance, physical consistency, generalization evaluation, trustworthiness and open benchmarks.
05Geophysical Agents
We explore geophysical agents that use language models to plan experiments, call specialist software and organize evidence. The first public case exposes SPECFEM 2D, 3D Cartesian and 3D Globe workflows as MCP tools, supporting automated and human-in-the-loop execution from parameter generation and meshing to solving and visualization; future work extends this pattern to processing, imaging, inversion and interpretation.
Featured project
SeismicBench
A shared coordinate system for seismic processing research
An open academic benchmark that brings seismic processing methods, datasets, papers and evaluation results together in a reproducible, traceable platform.
88Methods
44Benchmarks
368Papers
494Results
Verified 2026-08-28 · Source: project repository
Selected publications
Recent and selected work
Feature-Space Planes Searcher: A Universal Domain Adaptation Framework for Interpretability and Computational Efficiency
成葛马
GeoFormer: Geometry-Aware Transformer and its application to 5D First-Arrival Picking
高马
Robust Losses from Univariate Base Functions for Noisy-Label Learning
胡马
Towards Foundation Model-Driven Geophysical Inversion by Low-Rank Adaptation and Attention Gates
马
Foundation Models for Exploration Geophysics
刘马
Updates
Latest news
Benchmark data re-verified
Methods, benchmarks, papers and result records receive their August 2026 verification pass.
Learn more→SeismicBench v1.0 released
The first public release goes live across five core seismic processing tasks.
Learn more→SeismicBench project launched
The team begins building an open academic benchmark for seismic data processing.
Learn more→Academic contact
Open collaboration, grounded in real geophysical problems.
We welcome academic exchange across seismic processing, inversion, AI algorithms, foundation models and geophysical agents.
Public academic emailjwm@pku.edu.cnGeoBrain on Hugging Face ↗